Hi Roberto
Its a 15 consecutive frame pan in 24P, so this 15 frames amount to less than a second all up. However when you see it real time it is a reasonably slow pan, it takes about 9 seconds to get to the top of the metal post that holds the billboard, thats upwards...so its not entirely out of the frame (well within the slow pan recommendations)hower evn tho it's slowish, its also somewhat stuttery and jerky.(handheld) Even smoothed in Warp Stabilizer in CS6 it still looks kinda funky. I have done tons of smoothed shots but they always strobe...at 40 shutter and at 50 shutter, even after stabilizing them, better at 30 shutter tho, good...but definitely not perfect. If you note the dark car in the background (before the yellow sign)...moving at a moderate slower speed (30kh maybe), it has not moved that far thru those 15 stills. Also bear in mind that this is not an all Intra codec (its GOP6)... but all Intra does not look that much better (as far as panning is concerned).
hI @Astro
Not much movement there at all.. it's a kind of pan-tilt of just a few degrees and not much to go on. It's interesting to hear that there's stuttering with so little movement.
Tiny movements are much harder to measure than large ones, and inaccurate.
I was hoping for your typical Academy, 5-second pan side to side, then there would be more to measure with more accuracy. If you haven't got anything else I'll try it tomorrow using the spinning barrel I used for the jello tests.
@Roberto This may help somewhat too, its the clip those stills are taken from (scaled down to 32 meg) Its at 720P and lower bitrate (Gop6, 24P, Shutter 50...never higher than that) It has been smoothed in the CS6 Warp Stabilizer, so its improved on the original, but note that even tho its smoothed the bottom of the main billboard flickers and strobes as the camera pans up, and also the flicker-strobe of the white writing as the camera pans left ...almost jerking backwards and forwards (not so noticeable on the yellow writing tho). I know the later pans in this short clip are a little fast, but its the earlier part that is of interest. It starts approximately where those stills I sent you start. Here is the yousendit link https://www.yousendit.com/download/QlVpWmdndWM1R1AwZXNUQw
OK, I animated the frames into a GIF. It's not blurred and it stutters, of course. Fun but inconclusive, I'm afraid, because to see blur I have to magnify everything too much..
The attached close-up is waaay over-zoomed. I can't see any blurring but I'm not saying it isn't there.
The pana 25mm (super sharp) prime does amplify strobing, especially at deep depth of field and of course high contrast scenes. Have you guys tried the 80% variable mode? (I know, no sound in camera) Is it not 30p in a 24p wrapper or something? The goal being a smooth, regular framerate 24p clip rendered from the 80% setting. I tried it a bit, but can't be certain. Thanks for all your input here. If I have the opportunity to do some more scientific tests, I will post them here.
@Astro From information I gathered here on personal-view, it seems that a big problem is the sharpness that is always added in-camera to all micro 4/3 lenses, as even setting sharpness on theGH2 to -2 won't disable it completely. For this reason @LPowell suggests the use of a black diffusion filter with his Flow Motion, to reduce contrast in the same hi-con image parts where the in-camera sharpness is most active, without degrading the image too much. He suggests the Tiffen Black Diffusion FX3 filter, but other gradations are available (there's an entire thread here on that). Also Schneider makes some good ones. I bought a Tiffen Black Diffusion FX3 filter and it does indeed help with movement, even if it's completely useless on the lens I use the most, the Olympus 9-18, because the filter's pattern is visible in the frame as it's never completely out of focus. Now I'm looking at making a behind-the-lens net holder. It's not as easy as with PL mount lenses, but it might be feasible. Just wondering if anyone has gone that route already with m4/3 lenses...
Frankly speaking, I think lowering the contrast of the image going into the GH2 sensor just to control the choppy motion, might not be a good idea.
Too cumbersome an implementation, as well. If you have one single lens, sure it can be done. But I have a number of lenses, plus a number of adapters - as surely do many of us GH2 owners. It is not feasible to go Hurlbut http://goo.gl/jqICw on every lens, and mount/unmount that filter on location, ESPECIALLY on non-PL lenses...
I'd say, let the light in the way it is; but tehn deal with internal processing is the way that removes much of the annoying motion strobing. Which is: (maybe?) use Flow Motion hack by @LPowell, and (definitely) employ low shutter speeds of 1/25 or 1/30.
Very interesting thread! What's the negative of using lower ss in 24p mode like 1/25 or 1/30 in order to reduce sstrobing? More blur in anything moving in the frame?
@Astro You shouldnt be using Cluster v2 its defunct witht the wrong tables. Try all three new Cluster v4 settings (3 GOP CLuster, 6 GOP Cluster and 12/15 GOP CLuster) and report back. Then try the new Crossfire (GOP1 on 24p). Let me know how all of those 4 get on - I will take a serious look into this problem this weekend.
All these new settings can be found here;-
Thanks. Nick
@Driftwood Thanks heaps, Thanks for your input.
Like @mo7ies I agree that the lower shutter speed helps quite a bit in 24P. I did a lot of tests last night, panning close and slowly over a couple of guitars (incl a detailed Dobro)
Notwithstanding the Cluster V2 being out of date...I did like Quantam100 with all I frames (the detail of it) panning was kinda choppy tho (stroby) I had to have the shutter at 50 cause it was night here in Australia (PAL) but the detail was beautiful. Sedna was real nice too (again all Intra?) but slightly more demanding (I am using the 64/95Mb card) and harder to notice any real detail difference.
But on the subject of panning they were all a little strobey....However I will look forward to trying Cluster4/Crossfire etc.. I will try them today and post the results.
@rapsoe Yeah what you say is correct, the micro 4/3 lenses are sharpish for sure, but I have to say that the strobe issue does not go way even on the older Konica Lens like the 57mm, a strong ND filter does help tho, but does have its own set of problems IMO. One of the most telltale signs of what the GH2 does not handle well is in the slow pan over that Billboard I posted.
The white lettering over the black background pulses away whereas the yellow lettering is not too bad on the same pan...interesting, but the high contrast thing is something most people know about anyway, the difference is very clear on that clip tho, even more so on the unsmoothed 1080P version. Speak soon....this thread has helped a lot. Thanks a lot Cheers
Procedure:
Observation:
[Added] Table of threshold-processed images:
Note: The image grabs below are full-frame, un-retouched. Sorry they're so dark - (Daylight only, very overcast & GH2 was slightly closer than the lens's 2m minimum focus, with me standing at the edge of my veranda with a 1/2 metre drop behind me. I just needed to tighten aperture a bit so as not to fall over backwards!) However, we're not making pretty pictures here - just blur. All the shutter info is still intact.
GH2, Sanity 5 @ 194Mb, Zenit Zoom with a .75 Diopter, Set to 24P but I ended up getting 25 cos of hack, I guess.
None of this matters - yet. I was sort of expecting rogue shutter behaviour from the GH2 but it performed like a champ.
I'll do next tests [24P] using a no-hack, I think - firmware 1.1 (there's got to be a benchmark to start with).
I must say it's a pain getting anything to function at a sync speed these days; I used to be surrounded by turntables, crystal-synced Nagras & Uhers. Now I'm actually contemplating filming circles on paper as they leave my printer! Any ideas of stuff we all have, let me know pls!
Roberto>Set to 24P but I ended up getting 25 cos of hack, I guess.
If you mean 24p cinema? 24p stays 24p whether set to PAL or NTSC in a hacked camera. Only HBR, FHD and SH change regions. We would love real 25p thou.
I wonder if the test would be different with a white bin and black dot? ( better still a black vertical line)
It looks like you're new here. If you want to get involved, click one of these buttons!